Vision-based location sensing and self-updating information models for simulation-based building control strategies

Dynamic facility state models could effectively support simulation-based building systems control. In a simulation-based control strategy, the permutation space of control options (alternative states of control devices) at a future time step is proactively explored using computational simulation. The simulated implications of alternative control options are then compared based on users' preferences. From this comparison, the preferable control actions can be deduced. To achieve this functionality, however, the presence of a dynamic and self-updating building model (with context, room, systems, and occupancy data) is necessary. In this paper, we described the conception and implementation of a prototype visionbased object location sensing and occupancy detection system to provide the control unit of a sentient building with a steady flow of up-to-date building state infomration.